This Paper on NO2 photolysis reports an interesting method and its successful application. The topic fits well in with the scope of ACP. The paper is well structured and well written. It is concise, i.e., the authors do not waste pages for unnecessary information but; still, as far as I can judge, all necessary information is there, except for information on uncertainties. I like very much how the authors put their work in the context of existing work. Scientifically, the method applied appears sound to me. The study may not be world-shattering because largely existing knowledge is confirmed, but the  study increases confidence both in the modelling of J values and the measurements used.

That said, I must say that I am somewhat disappointed that the authors did not try to estimate the uncertainty of their inferred photolysis rates. In order to judge if this study adds incremental knowledge only or is a major step ahead, it would be necessary to contrast the estimated error of the inferred J-values with the estimated uncertainty of the modelled J values that are based on pre-exixting knowledge. I concede that a rigorous quantitative error assessment might be out of reach because the uncertainties of some ingoing quantities are not quite clear. But even a rough-and-ready uncertainty estimate would be far better than nothing. Or the authors could tackle this issue from the other side: They could estimate how accurate NO and NO2 measurements must be to allow a reasonable inference of J(NO2). 

Another issue I came across finally turned out to be much less dramatic than I first was afraid of. The MIPAS retrieval of NO uses a model that calculates the populations of the excited NO states. This model takes photolysis of NO2 into account, and uses the respective J values (See Funke et al., doi: 10.1029/2004JD005225, their R4). This suggests that there is the risk of a logical circle when MIPAS NO is used to infer the NO2 photolysis rates. To clarify this, I  contacted B. Funke, who gave the all-clear: Chemical excitation accounts only for 10 to 15% in the stratosphere, the rest is thermal. Thus, retrieved NO is only weakly to moderately dependent on the assumed J(NO2) values, and the use of MIPAS NO to confirm the modelled J(NO2) does not lead to a logical circle. However, in cases where the modelled J is totally wrong, the inference method presented by the authors might, strictly speaking, require an iterative approach where the revised J(NO2) is used for an improved MIPAS NO retrieval, and so forth. Since the authors confirm the modelled J(NO2) values, this is not an issue, but the method presented appears somehow contingent upon reasonable prior knowledge of J(NO2) to me. I wonder if it wouldn't be adequate to mention this issue in the paper. By the way: within the context of an unpublished sanity check of the MIPAS retrieval, B. Funke also inferred J(NO2) from MIPAS data. His results were consistent with the results of the paper under discussion. This increases confidence in the results presented.

A further concern is that in some cases the references are misleading because they do not always refer to the data actiually used. In summary, I recommend publication of the manuscript after fixing the issues listed below.

l22: strictly speaking, the sunlight entering the stratosphere depends on all absorbing species above. And, even more strictly speaking, as soon as surface-reflected sunlight plays a role, also concentrations of trace gases below may play a role. I suggest to weaken the statement accordingly, e.g. "... as well as the distributions of absorbing species, particularly the overhead concentrations of oxygen and ozone,..."

l31: I would prefer Greek letter "nu" instead of "v" for frequency (This may simply be an issue with the font chosen that does not allow to distinguish between nu and v)

l58: MIPAS does not measure backscattered radiance but infrared emission of the atmosphere. Just delete "backscatter". The fact that MIPAS measures limb emission is mentioned in the following sentence anyway.

l69: For the species under consideration, the vertical resolution reported here seems a bit optimistic to me. Funke (ACP 16 8667-8693, 2016) report a vertical resolution of 3-6 km for the sunlit stratosphere for NO; Funke et al., ACP 16, 8667-8693, 2016 report a vertical resolution of MIPAS NOy of 4-6 km. von Clarmann et al. (AMT 2, 1-17, 2009) report a vertical resolution of MIPAS ClO of 3.3-12.8 km for the altitude range 20-40 km. For ozone, the vertical resolution estimate seems realistic to me. However, the numbers quoted above refer to older data versions; I understand that version 8 data are used here; values might differ from those quoted above. But I think that the actual vertical resolution is reported along with the mixing ratios in the database and should be used here. Please check the actual values in the database.

l69: The horizontal resolution is 30 km across track. Along-track it is much coarser; for most species it is limited by the along-track sampling, which is appr. 500(400) km until(after) 2004.

l76: Since there exist multiple MIPAS data sets, it may be clearer to write "V8 level2 MIPAS IMK/IAA retrievals". Some of the references refer to older data versions or to another data product not used here.

l76/77: V8 NO errors are reported in Funke et al., 2022. This paper is meanwhile accepted for publication. The numbers (5-15% at altitudes between 20 and 40 km) are still valid, but the reference to Sheese at al. is obsolete because it refers to an older data version. 

l77: The NO2 random uncertainty reported in Wetzel et al. 2007 does not refer to the IMK/IAA data used here but to the ESA data product. Thus this reference is misleading in this context and should be removed. Unfortunately, no error estimates for IMK/IAA V8 NO2 data are available yet. Please see my suggestion below. V8 ozone uncertainties are reported by Kiefer et al. (2022). Thus, the reference to Laeng et al., which refers to an older IMK/IAA data version, is obsolete.

l78: Also von Clarmann et al. (2009), referenced here for ClO uncertainties, refers to an older data version. Unfortunately, also for IMK/IAA V8 ClO no journal paper exists yet.

l76-78: If the authors still want to acknowledge the work referenced (except for the obsolete references), I suggest something like "In this paper, NO, NO2, O3, ClO, temperature and pressure data from V8 MIPAS retrievals performed with the IMK/IAA level 2 processor were used. The retrieval of pressure and temperature is reported by Kiefer et al. (2021). The NO retrieval is documented by Funke et al., 2022. These athors report a precision of 5-15% for altitudes of 20 to 40 km. For O3, Kiefer et al (2022) report a precision of 2-5% in the altitude region of interest. The retrieval of NO2 and ClO is described in Funke et al. (doi:101029/2004JD005225, 2005) and von Clarmann et al. (2009), respectively, with precisions due to measurement noise of 0.2 - 0.3 ppbv for NO2 and more than 35 % for ClO but these papers refer to older data versions. Precision estimates for V8 ClO and NO2 are not yet available but the values quoted here can be used as a rough guideline." Since no quantitative use of the error estimates is made, all this does not harm the conclusions of the paper. But it should be avoided that the readers are misguided, thus my fussy comments in this context.

l76: A general remark on the use of error estimates: The authors refer to the precision. Isn't the accuracy or the total estimated error more relevant in the given context? I assume that any bias will affect the inferred J-values, while precision (random error) might not be too much of an issue as long as enough measurements are available.

l101: I have not quite understood if Eq. (2) is evaluated datapoint by datapoint or if any kind of averaging or regression is involved. If the latter is the case, the precision of the measurements would even be less relevant compared to any possible systematic error contributions.

l99-110: The retrieval of NO by Funke et al. involves a non-LTE model, which calculates the populations of the excited states of NO. This model involves NO2 photolysis and uses TUV photolysis rates. Please see my general comments above on this issue.

l169/170: A short explanation why NO2 photolysis is more albedo-dependent than that of other species would be nice. I suspect that this is because, as stated above, the atmosphere is quite transparent at frequencies relevant to NO2 photolysis, thus enough backscattered photons survive the long path through the atmosphere to Earth's surface and back to the air volume under assessment.

l200: I think that this statement might be a bit too strong or too general. If no J(NO2) data are available, then there is no reliable NO-retrieval from infrared emission measurements (see my discussion of this issue in the general part of this review). I would prefer some slightly weaker wording here. What the authors certainly can claim is that they present a new method to validate the modelling of J(NO2).

Conclusion: You may also wish to add a short remark that the nice agreement between modelled photolysis rates and those inferred from the measurements also increases confidence in the measurements used. Does the excellet fit between inferred and modelled J-values imply that the estimates of measurement errors of NO and NO2 are overly conservative?

The authors present a new method to infer the stratospheric photolysis rate of NO₂ using satellite (MIPAS) measurements. The photolysis rate coefficient determines the diurnal variation of NOx photochemistry. The results agree well with model predictions. This work provides the first observations-based validation of the role of albedo in driving polar photochemistry.

The scientific questions addressed by the paper are certainly within the scope of ACP. The title clearly reflects the contents of the paper, the abstract provides a concise and complete summary of the work, the authors provide proper credit to related work by other groups and clearly indicate their new contribution. The presentation is generally well structured and clear, and the language and mathematical notation is adequate.

My main concern is that the description of the approach is not sufficiently complete to allow their reproduction by others. For example, how do the authors collocate the model horizontal and vertical grid with those of the measurements? It is stated that “Model values for December 2009 at the same times and location as the satellite data are selected to compare with the satellite data”, but what are the actual spatial and temporal collocation criteria? Further, is any kind of interpolation (temporal, horizontal and/or vertical) done subsequently to match the grids up? This is critical information that is missing in the paper. Are aerosols or clouds considered in the comparisons? This is relevant because the authors use a four-stream radiative transfer model, which may not be accurate enough (especially in the UV/Visible spectral regions) when aerosols or clouds are present. The retrieval accuracies may also degrade in these scenarios.

The results presented by the authors agree very well with model estimates, but this begs the question: what is the use of satellite observations if models provide accurate results? The work does present an “observations-based check on the role of albedo in driving polar photochemistry”, but this result alone would only provide an incremental improvement to existing scientific understanding. It would be a lot more revealing if the authors could figure out under what conditions the models do not work so well (scenarios with aerosols and/or clouds?). This also leads to the issue of uncertainty quantification. There is no mention of error characteristics in the paper. This is critical for satellite-based retrievals. Without knowledge of the retrieval errors, it is very hard to make any evaluations about the quality and/or robustness of the results. For example, the statement that “However, in the stratosphere below about 33 km [O] has a small effect on JNO2 (less than 8.1 percent)” is meaningless unless it is contrasted with errors in JNO2 itself. The authors do report precisions for the various species. These could probably be used to obtain precisions for the photolysis rate estimates.

A few statements need references:

“However, in the stratosphere below about 33 km [O] has a small effect on JNO2 (less than 8.1 percent).”

“ClO can similarly be ignored when altitudes are less than 35 km, where ClO concentrations are small”

“HO2 and BrO both can react with NO but they are not measured by MIPAS and their contributions to the partitioning between NO and NO2 are negligibly small at the altitudes considered here.”

Overall, the paper has potential for publication after the changes listed above are made.